Multimodal Maestro is a Python package with utilities for prompting Large Multimodal Models (LMMs) such as GPT-4 with Vision.

Multimodal Maestro is launching with an implementation of Microsoft's Set of Mark prompting technique, outlined in the paper "Set-of-Mark Prompting Unleashes Extraordinary Visual Grounding in GPT-4" [1]. Using Multimodal Maestro, you can generate SoM masks with ease.

We are actively looking to add more utilities for LMM prompts. We would love ideas!

[1] https://arxiv.org/pdf/2310.11441.pdf